logs archiveIRC Archive / Freenode / #oracle / 2010 / February / 24 / 1
rizzo
idletask: what version?
you can't deactivate indexes unless it's 11g invisible indexes
making an index unusable is not a good idea. only real alternative is to drop it and rebuild it later
constraints you just run an ALTER statement (I believe ALTER TABLE foo DISABLE CONSTRAINT bar;)
idletask
I meant constraint, and yes, that's it
But it just looks like my nologging clause to alter table didn't make it :(
Or is that those pesky LOBs again?
spcshpopr8r
I'm getting horrendous performance of "merge into" operations between a table without compression and a table with "compress for all operations", can anybody refer me to performance improvements with "merge into" generally as well as specifically with regard to advanced compression
idletask
They do have separate storage clauses, don't they?
Aargh
f*ing LOBs
They're the bulk part, obviously
YESS
Now it works
Data import without having to go through the redo logs, _and_ lob deduplication
Now, time to sleep
Have fun
alpine_style
is there someplace you can go online to write SQL queries against a standard employee mock up schema?
you know, like there are these fancy pastebins and what not for perl, java, javascript, etc where you can test code on the spot, anything like that for SQL?
RJarett
alpine_style: just download oracle xe
its free and easy to install
alpine_style
RJarett: i'll try that, thanks
RJarett
http://www.oracle.com/technology/software/products/database/xe/index.html
alpine_style
yeah, it's loading now, a tad slow
lol
i just figured someone had ported this to a nice web app where you could just go and type some SQL and see what happened
of course with some visual schema browser, ah I ask for too much!
:-)
         

RJarett
pastebin sites arent emulators of the language you are running
alpine_style
RJarett: true indeed, i was thinking of something like http://jsbin.com/ for javascript which is an emulator
RJarett
javascript is a runtime languagewhich can be ran in your own browser
alpine_style
bummer, XE only for windows linux, me be on mac
RJarett
sucks to be you
alpine_style
i may fire up VMWare fusion and do it there
RJarett
or load up postgres
http://www.postgresql.org/download/macosx
alpine_style
is the employee schema pretty much universal across all the RDMSs?
RJarett
no
plus sql isnt universal
alpine_style
hmm, i thought most of the "modern" RDMS where ANSI SQL compliant, no?
RJarett
ansi doesnt mean standard
theres levels of ansi
different versions based on acceptance date too
alpine_style
sure, it's complicated, welcome to CS
lol
RJarett: thanks for you kind help, take care
off the check on the Olympics
samuel_
hello guys
RJarett
hola
samuel
so, does anyone know when APEX 4 will be released?
I am looking forward to websheets
thepro4ever
what is the oracle mt recovery service used for? i installed the oracle client and for some reason the install created that service for me
idletask
Hello
goschtl
hi i am new to oracle, i try to add a primary key constraint to a view. alter table unternehmen add constraint mnr_pk PRIMARY KEY (MNR); but i get this error... *
ERROR at line 1:
ORA-00942: table or view does not exist
but with desc unternehmen i see the view?
idletask
goschtl: you cannot put primary keys on views, and this is not specific to Oracle
You can put primary keys on tables, that's all
goschtl
idletask: ok thanks for info
         

sidh
Greetings Gentlemen
idletask
Hello
goschtl
is it possible to create a table with a ForeignKey Constraint to a table from a database link. I don't get it to work?
idletask
No, it's not possible
Well, I don't think it is, and if it is, it's a very, very bad idea to do so (think "insert performance)
goschtl
idletask: so what is the best way todo it. export import, any kind or replication?
idletask
The best thing would be to have both tables on the same instance to start with :p
goschtl
idletask: unfortunatly this is not possible
idletask
goschtl: then the design is fundamentally broken
You'll never be able to obey that fk constraint reliably
goschtl
idletask: ok i have to think again and harder on this thanks for your help
tijnie
goschtl, what you can try is create a trigger. This trigger can check whether there is a "corresponding: row on the other side of the line.
But I believe I have to agree with idletask : I don't like the design :-)
Only think of network outages....
hali
mviews perhaps
idletask
Question again wrt imports
I disable logging for tables and lobs, disable constraints, import data, re-enable logging and constraints
When I reenable constraints, do I also need to rebuild the indexes?
Gee, I'm such a newbie
If I do a content=metadata_only, this only concerns tables, right? Not things such as procedures and the like?
tijnie
idletask, the index will be build when the data is inserted into the tables. Indexes point to physical locations in datafiles. so literally importing them has no use, because the the rows the index points to are defenitly on another position. Hence the index must be created while the data is submitted.
hali
idletask: people generally rebuild too often, there are procs to check index status'
checking index leaf depth etc
superlinux
hi guys
how do you dump a CSV file into an oracle table?
usn_work
sqlloader
superlinux
sqlloader? what is it exactly? i mean how do i get that?
dsdel
superlinux
ok
thanks
usn_work
don't be afraid to use a search engine ;)
superlinux
usn_work, i did use.
but i always enlarge my pool of search
better probability
usn_work
Yeah, no trouble asking here like you did, that's what we are here for. But looking up a given keyword would be nice
idletask
Aaah, google
But anyway
Asking here is more to the point ;)
hali, tijnie: yes, but I _disable_ constraints while importing data, for performance reasons - or is this a bad idea, given that I don't log anything while importing anyway but only re-enable logging afterwards?
hali
should be ok
idletask
Hmm
(Action) needs to wrap his head around
tijnie
idletask, not logging is a fine idea. No problems there. If you are importing a "complete" database or schema I don't think diabling keys and later enabling them does make a difference.
fuse405
does anyone know a site where I can order/get oracle 9i release 2 on cd-rom/s?
tijnie
fuse405, can't you download them, and then burn them yourself?
idletask
tijnie: that's a complete schema here
tijnie
idletask, If you dit it in one pass, you don't have to dis/enable constraints.
idletask, If you first import the meta-data (ddl) and in a second pass the data then you probably should yes.
Just out of curiousity: If I create an index with "pctfree 99" is this only for the leaf blocks or also for the branch blocks?
idletask
tijnie: I import the metadata and then the data, yes, because the data contains lobs which amount for a huge volume (185 GB for the highest) which I obviously don't want to see go through the redo/archive logs :p
tijnie
:-) I understand. Then you have to disable constraints. (I believe youanswered following allready earier but:...can't you stop archive log mode for the database, and afterwards put the database in archivelog mode again?)
RJarett
tijnie: in 10 and 11 dont specify those options anymore
idletask
tijnie: no I can't
Because this requires bringing the instance down, which I cannot afford when this will go into production
So, tables/lobs nologgin during _data_ import is the only solution
RJarett
you can specify pctfree with assm but it isnt used very much. if you do use them, make sure your calculations are correct and dont use old logic of pre assm days
MaL0
hi
there's any parameter on Oracle application server that define's the client timeout ?
RJarett
MaL0: thats within apache if you are using that as your listener
plus within your servers OS itself as to how it handled clients in time_wait
MaL0
ok, thank you RJarett
RJarett
TimeOut in apache
but it should match or be near the servers net.ipv4.tcp_fin_timeout setting
or else the client browser may throw an error if the server times out before apache
MaL0
oks =)
tijnie
RJarett, hahahaha I almost choked...I thought that your remark saying " in 10 and 11 dont specify those options anymore" was talking about the question to idletask whether he could sitch off archivelog mode.
RJarett
no to your pctfree
tijnie
Yup....I realized that after a while :-)
I was just playing around with pctfree (pctused can not be used in idexes). My database are in assm. But it's never to late to learn some background.
RJarett
db sizing used to take a ton of thought and effort
nowdays no one thinks about it
tijnie
RJarett, I'm not quite sure whether that is positive.
RJarett
the new fad is partition alignment
tijnie
(not thinking often means: no understanding. For a technician that is a sad thing)
idletask
So, as to the other question... impdp's, or expdp's data_option parameter is relevant only for tables/indexes, or is it also relevant for, say, procedures (ie you get the definition but not the text of it if you specify metadata_only)?
tijnie
idletask, I believe that option is only relevant for tables
not for procedures/functions/etc.
idletask
tijnie: OK, thanks, that's what I thought as well but I wanted another point of view on that particular point
tijnie
No problem, but honesty requires me to tell, thaht I didn't try it. I think it works that way (just as the docu. thinks it will wotk so)
wotk==work
hydester
anybody using the PHP rpms from oracle? do you know if PCNTL is enabled on them?
idletask
Uh, why on earth would you use PHP RPMs from oracle?
mator
:)
idletask
Their already existing RPMs are bad enough, but at least they limit themselves to their own products
RJarett
because they are ZEND certified and have some little extras in there
mator
"php -i" on oracle installed php rpms will show does it have pcntl enabled or not
hydester, why to bother?
hydester
for some reason i thought using the oracle rpms avoided having to compile in the oci libraries
i thought using rpms was cleaner for configuration management. is this wrong?
mator: what method do you recommend?
idletask
hydester: RPMs are made by human beings, and human beings are prone to errors
But "official" humans (ie, the ones who work for distributors) less than others
hydester
idletask: does not using RPMs throw away the benefits of dependency management?
usn_work
(Action) is an inofficial human
mator
hydester, well, ofcourse rpm is usefull at software management (not configuration), but as php is heavily modular it can be compiled almost any time
hydester, i'm not recommending anything, since it up to you what approach to get
steelnwool
what is the general opinion of the Unbreakable Linux offering?
mator
steelnwool, i really don't know anyone running oracle linux, most of people running rhel5/centos or some other oracle certified OS/linux distrobution
RJarett
bastardized redhat with slower updates?
usn_work
To have a single point of support
idletask
hydester: sure does
steelnwool
motor/rjaret, thaose were my guesses.
idletask
hydester: but, as an RPM professional for more than 10 years, I can tell you that "out of distribution" RPM makers rarely do a good job
RJarett
centos 5
steelnwool
i thought "jeez whiz, a really small community, that sounds.... great"
idletask
hydester: one notable exception is Dag Wieers, this guy does a wonderful job, and if you want out of tree RPMs, this is where you should look at
dag.wieers.com
RJarett
hes been pushed to rpmforge
thats the only external repo you really need
idletask
RJarett: I quite agree - that, and jpackage, which does a very good job too
(if you use anything Java)
RJarett
nope
well we do, but its all our homegrown stuff
so we are very very protective of changes on the system
idletask
So are we
But we have trust sources
Dag Wieers is one, JPackage is the other
(and my own RPM packages as well, heh)
hydester
idletask: so in summary, you recommend these other rpm repos than fedora/redhat/centos (if using corresponding OS)?
RJarett
yes
but use the oracle php rpm if youll have heavy load
hydester
define heavy load?
RJarett
more than a few concurrent sessions
or heavy ajaxification or JSON or comet
hydester
because the OS repos and compiling doesn't have good defaults set?
or DB-related
RJarett
because non oracle php makes hackish OCI calls out to the libs
square peg round hole type passthrough of calls
hydester
ok, so in my specific issue this is related to a server CLI script. so then probably use oracle RPMs on web server and compile to enable PCNTL on server that runs cron php scripts?
RJarett
so you JUST need php-cli to have this?
or will the webserver be calling to the host
thats how i would do it. id have my own env for the php-cli
hydester
i just need php-cli to support pcntl, which only would work for cli anyway
right, but it needs to be enabled via compile flag or some rpm with it enabled. i wasn't aware of it being an ini toggle
Process Control support in PHP is not enabled by default. You have to compile the CGI or CLI version of PHP with --enable-pcntl configuration option when compiling PHP to enable Process Control support.
i just know i have some machines with it working on, and some where it doesn't. none of those were compiled. so i was determining the rpms used, etc. and trying to figure out why there is this difference
idletask
Well, Know Thy Enemy
It's true for everything*
I don't do PHP :p
hydester
i wouldn't want to blame PHP for this. why not blame redhat or the rpm repo?
« prev 1 2 next »